2

When looking around for examples of how five levels of defect severity are named it turned out that almost everybody seem to use four levels of severety (they name them differently though).

Is there a rational reason for this? Any fact that says that this is actually the best thing to do?

Or is it just that everybody does this because everybody else does it? Which means that it's a reasonable idea to (go against the flow and) use five levels instead.

As I mentioned I've got no great idea what to call these five levels (perhaps critical, major, medium, minor, trivial or something). The idea is that the highest of course is catastrophic/critical/blocker (ie something that we simply can't stand or nearly so) then the lower would be geometrically smaller in expected impact. Where it ends is when the expected impact is too low to bother about.

  • 1
    Hi skying, welcome to PMSE! I believe it's a very good question, as I never thought there'd be other ways of seeing severity (beyond the common four, example in softwaretestinghelp.com/…: blocker / major / moderate / minor)... What would be your 5th category? I believe adding this up to your question can add value on it. – Tiago Cardoso Nov 29 '17 at 8:40
  • Because 4 boxes typically make up a magic quadrant. You can assign 127 different levels of severity in your organization if you really want to, and if you can rationalize a meaningful distinction between them, but studies show people cognitively intuit only up to 3. And so: What framework are you using? Why do you feel you need a different scale? Why can't you just add another level to your current system? – Todd A. Jacobs Nov 29 '17 at 17:10
  • 2
    @ToddA.Jacobs "Because 4 boxes typically make up a magic quadrant": that looks like irrelevant justification or number magic. "studies show people cognitively intuit only up to 3": that looks more like a reason, which studies do you refer to? Technically I can set an arbitrary number of levels, but the question was if there's a founded reason why just four levels is so widely chosen. – skyking Nov 30 '17 at 6:05
  • @skyking +1 (or more) – Alan Larimer Dec 1 '17 at 15:33
1

I doubt there is any absolute rational. However let's look at it from a different perspective. Let's agree that there are showstopper defects which must be corrected before the current version of the product can be released. On the other end of the spectrum there are trivial defects which don't really impact the product and may never get fixed, such as a verb tense being wrong in some text.

So we are left with how many types of defects are left in the middle. Well there are those that we should fix in the next version, and those which which might fix in some unknown future version.

So pragmatically (whatever you call them...)

Severity 1 - Must be fixed before this version is ready.

Severity 2 - Should be fixed in next version

Severity 3 - Should/Could be fixed in some future undetermined version

Severity 4 - Trivial defect which may never get fixed

Now depending on the product and how many versions are planned you might want severities for current version, 2nd version, and the 3rd version as well as unknown version. So you'd need 5 severities.

It obviously doesn't help to have 500 severities because you'll waste an inordinate amount of time twiddling severity numbers.

-1

The norm of following four levels of Severity and Priority comes from consensus and is driven by social psychology.

  • 1 category does not exist.
  • 2 categories is too broad. (Major, Minor)
  • 3 categories leads to debates. (Critical, Major, Minor)
  • 4 categories is just right. (Critical, Major, Minor, Cosmetic)
  • 5 categories is enough. (Blocker, Critical, Major, Minor, Cosmetic)
  • 6 categories is overkill. (Blocker, Critical, Major, Significant, Minor, Cosmetic)
  • 7 categories! (What are you doing? Creating a rainbow?!)

A major debate is usually between 3, 4 and 5 categories. Depending on the size and spread of the application and it's functionalities; the categorization can and should vary.

I believe 4 is just the right number of levels for majority of the applications and the 5th level should be added when there is slight overlap between critical & major defects. For example; large social media websites can implement a feature for a global audience but unfortunately it does not work for a particular geographical area/cell phone vendor/mobile device/version of the app.

Such defects even though they result in data/experience/reach loss, do not impact the majority audience and the impact is comparatively / relatively less than what is considered as blocker within the organization or for the application. However it needs to be addressed immediately to avoid any bad feedback/outrage, etc. and hence is critical. Categorizing such defects into a new level of impact enables better prioritizing/scheduling/manageability. After all enabling efficient management is primary purpose of categorization into priority & severity.

So in a nutshell, in my opinion from my experience, the size and reach of the organization / application should decide if you should switch from 4 levels to 5 levels of defect severity.

  • I see this answer (although strongly opinion based instead of evidence based) is pretty ok, would like to know why the downvote... – Tiago Cardoso Dec 3 '17 at 12:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.